Philosophical Issues, 15, Normativity, 2005
ERROR THEORY AND THE POSSIBILITY OF NORMATIVE ETHICS
Russ Shafer-Landau University of Wisconsin, Madison
There is a worry about morality that is as old as philosophy itself— namely, that morality is make-believe, an artifact of power relations, insecurities, and timid thought. The moral rules do their work as devices of social control and coordination. But these rules cannot be what they purport to be. Genuine moral rules are those that impose demands on us regardless of our desires or interests. Their authority does not stem from any instrumental link to what we care about. We are morally bound to aid the vulnerable, or fulfill our promises, even if doing so is nothing to us. The perennial worry is just this—that there is nothing underwriting such demands. Morality is a fiction. If this worry can be vindicated, then obviously the prospects for developing a normative ethic are bleak indeed. If there are no truths within morality—only a truth about morality, namely, that its edicts are uniformly untrue—then the enterprise of normative ethics is philosophically bankrupt. Normative ethics is meant to identify the conditions under which actions are morally right, and motives morally good or admirable. If nothing is ever morally right or good, then normative ethics loses its point. I do not believe that morality is a fiction. I think instead that morality is objective in a very strong sense. I don’t intend to argue for that here,1 but rather will limit myself to an effort to display why the strongest extant arguments for the fictive view are weaker than they might appear. Doing this will not provide us with the conditions under which actions are morally right, or motives morally good. It will not provide us with concrete normative ethical answers. Rather, it will, if successful, remove the most serious obstacle to the very enterprise of normative ethics. If my arguments are on target, then we may, with clear conscience, proceed in our normative ethical inquiries with the hope that has animated moral thinkers for
108 Russ Shafer-Landau
millennia—namely, that of lighting on the truth about the nature of what is truly right and good.
The Error Theory The ages-old worry about morality, that it is nothing but a convenient fiction, supplies the makings of an error theory of morality. Such a view assigns some conceptual presuppositions to moral talk, and claims that these presuppositions cannot be vindicated. Error theories can be distinguished from one another on the basis of the presuppositions they assign, and criticize. To sustain the charge of error, then, we must first be sure of our target. Richard Joyce, in the course of elaborating what is surely the most elegantly written, comprehensive and well-argued defense of a moral error theory yet to appear, introduces us to the notion of the conceptually nonnegotiable.2 Conceptually non-negotiable propositions are those that are embedded within theories, and have a special role to play therein. They are conceptually necessary conditions that are such that their falsity undermines the entire theoretical edifice that has been raised on their foundations. To be non-negotiable, then, is to play the role of a conceptual linchpin. It is a non-negotiable aspect of being a unicorn, for instance, that the thing in question resemble a horse and have a single horn emerging from its forehead. Anything that fails to meet these requirements cannot qualify as a unicorn. Since nothing does, in fact, meet these requirements, we are entitled to be error theorists regarding any sincere assertions that unicorns exist. Joyce traces a similar line of reasoning in defense of his moral error theory. The master argument he advances is simplicity itself. Its first step identifies what is conceptually non-negotiable about morality, and the second argues that this necessary element fails to obtain. It follows that morality itself is a fiction. Positive moral claims have the same status as those offered by nineteenth century phlogiston theorists, or ancient worshipers of sun gods. They are uniformly untrue. What is the non-negotiable element of morality? Joyce identifies it as categorical reason-giving power. Necessarily, moral demands supply agents with reasons, and this quite independently of whether the moral demand is at all conducive to satisfying one’s interests, or securing one’s wants or needs. This is meant to be a conceptual claim: the very notion of a moral demand carries with it the assumption that such demands impose on agents a reason to act regardless of whether doing so furthers their ends. Call such reasons categorical reasons. Joyce does not believe that they exist. Since morality depends on such reasons, moral claims are never true. By itself, this conclusion is compatible with either an error theory, or
Error Theory and Normative Ethics
various forms of noncognitivism. Whether we have fallen into error depends on the nature of our non-negotiable assumptions about morality. Unlike recent expressivist efforts, which hope to accommodate our common sense views about the moral realm, Joyce thinks that our ordinary understanding of morality contains within it a crucial assumption that cannot be vindicated.3 This core assumption—that moral duties generate categorical reasons—is false. Hence the error. Because there are two steps to Joyce’s argument, there are two points of possible criticism. One can take issue with his characterization of what it is about morality that is conceptually non-negotiable. Or one can grant him that, but reject his criticisms of categorical reasons. My strategy will be to pursue both challenges. Morality’s fate does not hinge on the existence of categorical reasons. And even if it did, the arguments that Joyce offers against such reasons are not compelling.
The Concept of Morality Distinguish two claims: (A) Morality is categorically applicable: a person can be morally obligated to do something even if doing it fails to serve any of her interests, wants, or needs. (B) Morality is categorically reason-giving: a person has a reason to do her moral duty even if doing her duty fails to serve any of her interests, wants, or needs.
(A) is a claim about the potential content of our moral duties—they are not limited to actions that serve the ends of the agent under obligation. (B) is a claim not about the content of such duties, but rather about their power to generate reasons for conformity. Joyce thinks that we’d class a theory as a moral theory only if it incorporates a commitment to (A) and (B). I am willing (tentatively) to sign on to the claim about (A). We might say that there is a kind of inescapability built into the very idea of morality—one’s moral duties do not depend for their existence on a showing that they are ancillary to one’s ends. The dedicated evildoer isn’t freed of his moral duties just because good conduct brings him no rewards. I confess to some reservations here. There are ethical subjectivists, relativists and egoists, after all. The deepest motivations for such views are often traceable to a specific combination of widely held claims: first, that moral duties entail reasons for action, and, second, that there are such reasons if and only if the actions they advise further an agent’s ends. (Joyce calls this latter claim practical instrumentalism, and fully endorses
110 Russ Shafer-Landau
it.) The resulting view has it that our moral duties, like our reasons, must find a source in an agent’s ends. It doesn’t seem like an incoherent view. But if one thinks that (A) is non-negotiable, then one must regard this line of reasoning as conceptually confused. If categorical applicability is built into our idea of morality, then views denying this kind of applicability cannot qualify as moral ones. We used to hear this about ethical egoism, and perhaps such critics were right.4 I’m not confident of a verdict here, because I am unsure about the conceptual boundaries of the moral realm. Let’s proceed, however, on the assumption that tinkering with (A) is changing the subject, and moving us to non-moral territory. If that is so, then morality can be vindicated only if there are moral duties that apply to us in a categorical fashion. Joyce offers us a master argument against the existence of such duties (and so, by his lights, against all moral duties): 1. If there are moral duties, then they are categorically applicable. 2. If there are categorically applicable moral duties, then these supply categorical reasons for obedience. 3. There are no such reasons. 4. Therefore there are no categorically applicable moral duties. 5. Therefore there are no moral duties.5
If we are prepared to grant (A), then premise (1) follows straightforwardly. Let’s postpone a discussion of (3) until the next section. What can be said on behalf of (2)? At this point Joyce relies on what he calls Mackie’s platitude: it is necessary and a priori that, for any agent x, if x ought to F, then x has a reason to F. (38) Mackie’s platitude is supposed to range over all normative domains, and the qualifications on any given ought transfer to those of its correlative reason. So a legal ought entails a legal reason; a prudential ought entails a prudential reason; a categorical ought entails a categorical reason. But there is good reason to doubt whether Mackie’s thesis is a platitude at all. Platitudes are claims that, at the very least, are obvious to those who understand them. Such claims are prime candidates for the conceptually non-negotiable; those who deny genuine platitudes invariably end up changing the subject, or entangling themselves in merely verbal disputes. It is a platitude, for instance, that red objects are colored objects. Anyone who denies such a thing means by ‘‘red,’’ or ‘‘colored,’’ something different from the rest of us. To us, it’s just obvious that the claim is correct. Mackie’s claim is not like that. Suppose Joyce is right in regarding claim (A), above, as conceptually non-negotiable. Thus people see morality as imposing duties whose fulfilment can, in cases, only frustrate desire, and stifle self-interest. Those who query whether there is reason to undertake such sacrifice don’t seem to be engaged in verbal trickery. Knowing that doing your duty will bring you
Error Theory and Normative Ethics
only misery, you reasonably ask: why do it? It’s possible, as Joyce alleges, that there’s no good answer to this question. But surely it is a sensible one to ask. Joyce rightly takes an article by Philippa Foot as a touchstone in the relevant debates in this area.6 Foot argued that morality does purport to be categorically applicable. But it is not, for all philosophers have managed to show, categorically reason-giving. She wasn’t an error theorist, however, because she rejected Mackie’s platitude, and with it, the claim that the very idea of a moral duty entails that of a power to generate categorical reasons. If Joyce is correct, then this forces us to say that Foot has changed the subject—she’s no longer speaking of morality. That surely is a verdict of last resort. What Foot offered us was a view that combined two theses: that of the categorical applicability of morality, and practical instrumentalism. Morality is relevantly ends-independent; our reasons, by contrast, must bear a link to what we care about; therefore the demands of morality and those of reason can come apart. This is not an idiosyncratic view. Consider the roster of those who regard it as an open question as to whether we necessarily have reason to do our moral duty: Mill and Sidgwick, and more recently Gregory Kavka,7 David Brink,8 Peter Railton,9 and David Copp,10 to name only a few, have allowed for the possibility that we may lack reason to do our duty. Indeed, one way to view the entire history of moral theorizing is as a series of efforts aimed at answering the amoralist—the one who acknowledges something as his categorically applicable moral duty, but remains indifferent and sees no reason to comply. The puzzlement that competent speakers sometimes feel when confronted with what they take to be their duty, and the many philosophical efforts to address the threat of amoralism, are, together, excellent evidence that Mackie’s platitude is no platitude at all. There is one way I can think of to vindicate Mackie’s platitude in the face of these objections. And that is to restrict its scope to all-thingsconsidered oughts—the sorts of things that summarize an overall normative verdict for a case and represent the in toto balance of normative considerations. Not every ought would obviously entail a reason—just all-thingsconsidered oughts. Even if we accept this amendment, this gets Joyce what he needs only if it is a platitude that moral oughts are invariably all-things-considered oughts. And that is very doubtful. Many deny the truth of such a claim, and that is enough to demote it from the category of the platitudinous. Those in this camp will argue that moral duties are not invariably maximally stringent, that they are sometimes best set aside for, and outweighed by, nonmoral considerations. Of course there are those who insist that moral oughts always receive normative priority, but the depth of the disagreement here is excellent evidence that, whichever side is right, it’s not a conceptual constraint on intelligible talk about our moral duties that they
112 Russ Shafer-Landau
invariably specify all-things-considered oughts. If that is correct, then even if all-things-considered oughts do entail practical reasons, Joyce needs an argument to show that moral oughts are, conceptually, all-things-considered oughts. That argument isn’t anywhere to be found. I’m not saying that such an argument can’t be made. My aim is much more modest. I have tried to undermine the only support that Joyce offers for the second premise of his master argument. That premise tells us that categorically applicable oughts entail categorical reasons. Why believe that? Because Mackie’s platitude entails it. Why believe Mackie’s platitude? Because it’s a platitude! But it isn’t. And therefore Joyce needs either to defend Mackie’s platitude, or to identify a different basis for supporting his second premise. In sum, even if we grant that claim (A) is a conceptually non-negotiable aspect of morality, we as yet have no reason to attribute such status to (B)— the claim that moral duties entail categorical reasons for action. And so, even if Joyce were able to show that there are no such reasons, this needn’t spell the demise of morality. The fate of morality does not hinge on the existence of categorical moral reasons.
The Arguments against Categorical Reasons Suppose the arguments I’ve just offered are unsound. Then we’d have to accept that categorical reasons are a nonnegotiable element of any credible morality. In that case, Joyce’s master argument would be vindicated if, and only if, there are no such reasons. At this point I want to turn my attention to Joyce’s arguments for denying that any such reasons exist. On my reading, he gives us two arguments designed to establish this conclusion.11 The first is implied by a number of passages that appear in the second chapter of Joyce’s book. Consider the following: (a) ‘‘Whenever we tell someone that she ought to do something, the question ‘Why?’ is perfectly legitimate. But we want to say something more than ‘Well, you simply must, and that’s all there is to it!’ We will want to provide her with a reason.’’ (39) (b) When speaking to a felon, ‘‘We say (at the very least) ‘You ought not to have done that.’ We cannot end matters there, or we have nothing with which to counter the felon’s ‘So what?’ Indeed, if all we had to say on the matter was ‘You simply mustn’t!’—accompanied by some table pounding— the felon’s query seems positively reasonable. We seek something that might engage the criminal. Even if it is something that does not succeed in actually persuading her, we want something the ignoring of which would be in some manner illegitimate on her part.’’ (44, Joyce’s italics)
Error Theory and Normative Ethics
(g) Joyce also introduces us to a very odd game, in which participants anticipate the actions of a stranger and give points on the basis of their predictions. There are rules to the game, and participants secretly issue ought judgments to ‘‘their’’ strangers, silently encouraging them to act in various ways. Yet, certainly, we want to agree with Joyce and say that ‘‘the rules and reasons of the game need not influence his [the stranger’s] decision procedure, and may quite reasonably be ignored.’’(41) These passages provide us with three related constraints on what can qualify as a reason. (a) tells us that (C1) A reason is something that answers normative questions; ‘‘You just should’’ is never an acceptable answer to such a question.
(b) implies that (C2) A reason is something that must be able to provide an agent with a rationale for action that she can understand and accept.
(g) states that (C3) If one has a reason to F, then one can’t reasonably ignore F.
Let’s summarize these constraints by saying that, necessarily, reasons must be capable of engaging the agent whose reasons they are. And now we have the makings of a master argument against categorical reasons: 1. If there are categorical reasons, then, for any reasonable agent S, S might have reason to F, but fail to be engaged by F. 2. Reasons cannot fail in this way. 3. Therefore there are no categorical reasons.
Premise (2) summarizes the constraints on reasons that Joyce has endorsed, above. Premise (1) is meant to follow from the nature of categorical reasons, and the nature of engagement as here understood. To engage an agent in this sense is to motivate her, to spark an interest, to provide an affirmative answer to the question of a consideration’s importance. Let us grant that, in order to do any of these things, an action must be (or be believed to be) ancillary to an agent’s ends, broadly understood to encompass her interests, wants, projects or needs. Yet categorical reasons are supposed to obtain independently of their relation to an agent’s ends. Certainly, a categorical reason might, as a contingent matter, nicely line up with the ends we happen to have. But it’s of the nature of such reasons that such a connection is
114 Russ Shafer-Landau
inessential to their existence. This argument purports to show that precisely such a connection is necessary for something to qualify as a reason. If it is, then there are no categorical reasons. My strategy for considering this argument is to make a concession, and then to criticize. I will concede premise (1). Friends of categorical reasons might resist it, by importing constraints on reasonableness that ensure that all reasonable people will be engaged by any categorical reason that applies to them. The question here is a complicated one, and requires for its answer an extensive foray into the nature of motivation, personal ends, and what constitutes the closing of a question for reasonable people. I’ll sidestep these important issues here, and proceed on the assumption that a plausible reading of (1) will give us a truth. That’s the concession. And now to criticism, which will be directed to premise (2). Because this premise is shorthand for the three constraints (C1)-(C3), it inherits their (im)plausibility. I see each of these constraints, and the passages that contain them, as question-begging. My conclusion, then, will be that premise (2) has yet to receive any adequate defense. Until it does, we haven’t been given a compelling argument for doubting the existence of categorical reasons. These constraints amount to the claim that something is a reason for a person only if that person can’t reasonably ignore or repudiate it. When is such a condition satisfied? Joyce doesn’t say, exactly, but we can get a sense of his intentions from the passages quoted above. I think it’s fair to extract the following criterion: a person can reasonably reject or remain indifferent to a putative reason just in case this claim on his attention is one that bears no instrumental link to his ends. We’d be unable to convince a perfectly reasonable agent that he has a reason to do something, if doing it bore no relation to his commitments. In such a case, the proposed action is nothing to him. Insisting that it must be done is nothing other than table pounding—pragmatically effective, sometimes, but entirely irrelevant to establishing the existence of genuine reasons. The problem is that there is no argument offered for this constraint. And some argument is needed, for the claim is subject to serious doubts. As a conceptual matter, a reason is a consideration that makes something more justified, legitimate or appropriate than it otherwise would be. (I’m not trying for a reductive definition here, just a truism that all parties to debates about reasons should accept.) Some argument is required to get us from this concept of a reason, to the specific requirement that any such consideration be capable of figuring in a justification acceptable to a hearer who is perfectly informed of his commitments, and can reason flawlessly with them as starting points. It’s possible that no amount of clear thinking, starting from one’s deepest commitments, will get one to appreciate the force of certain considerations. And that is because their normative force does not depend on their bearing an instrumental link to one’s existing commitments.
Error Theory and Normative Ethics
Consider a Thrasymachus purged of false factual beliefs and perfectly efficient in identifying and securing means to his chosen ends. Imagine that a very slight effort on his part will save a young girl from a miserable death. Given his starting points, he sees no reason to offer such aid. And nothing we could say, appealing only to nonnormative facts and logic, would get him to come over to our side. Therefore, according to Joyce, he has no genuine reason to help. There is nothing about the child’s situation and the relative ease with which he can help that favors his doing so. It all depends on the contingent connection (or its absence) between his ends and her life. But it seems, on the contrary, that there is excellent reason for him to save the child. There is something to be said for his saving the girl, even if Thrasymachus can’t see it. That, of course, is just assertion, not argument. But the assertion comports with common sense, and shows, at the least, that some quite strong argument is needed to vindicate the constraints that emerge from the passages I’ve quoted. The constraints are not self-evident. On the contrary, they assume that the content of reasons will be transparent to, and will silence the questioning of, those who reason perfectly from their given commitments. And this assumption is true only if something’s being a reason is conditioned on its bearing some instrumental link to our existing commitments. In other words, the constraints express a commitment to the truth of practical instrumentalism. But one can’t simply assume its truth in an argument designed to impugn categorical reasons. That is a petitio. So let us try another argument (which appears at 108ff.). The second one goes as follows: 1. An agent has a reason to act only if her awareness of that reason can explain why she acts as she does. 2. An agent’s awareness of a putative categorical reason cannot explain her actions. 3. Therefore there are no categorical reasons.
Let’s grant (2), in its most plausible form. An awareness of a putatively categorical reason can explain action, if the content of the reason is, or is believed to be, identical to that of an instrumental reason. But imagine instead a scenario in which an agent is confronted with a putative reason whose link to her concerns is indiscernible. If, as seems plausible, she will not endorse such a reason, then it cannot factor into her deliberations in such a way as to explain her actions. But why should this entail the absence of a reason? Why, in other words, should we think that premise (1) is true? Joyce does have an answer. If reasons were unable to explain action, then agents could be deeply alienated from their reasons. And this is impossible. So reasons must be capable of explaining action.12
116 Russ Shafer-Landau
Obviously the plausibility of this line depends on the claim about reasons and alienation. Let’s say that a reason explains an agent’s action just in case an agent’s awareness of it motivates her (at least to some extent) to undertake the action. Why must reasons motivate in this way? Because ‘‘If a normative reason could not potentially motivate an agent, then, if presented with such a reason, an agent could say ‘Yes, I accept that is a normative reason for me, but so what?’—and this, I have urged, is unacceptable.’’ (108) Joyce may well be right to judge such a response unacceptable. His argument for its unacceptability is excellent. (It appears in his book on pp. 49– 51.) The problem is that a reason’s motivational failure—and so its inability to figure in an explanation of an agent’s action—does not thereby license this sort of unacceptable response. If a reason is incapable of motivating an agent, he will reply to its presentation with indifference. But he will also, almost certainly, deny that the consideration is a genuine reason. He surely need not have the incoherent response of acknowledging a reason and denying that it has any bearing on his deliberations. We can see this quite clearly if we distinguish two kinds of alienation that can result from a thorough consideration of one’s commitments and the implications that follow from them. In both cases, one is left cold by the putative reason. In the first case, the agent admits that she has a reason, and says so what? Joyce claims that such a response is incoherent, and, as I’ve said, he may well be right.13 In the second case, the agent’s indifference is accompanied by a denial that she has a reason. But this denial can have two explanations. The first has it that since the putative reason does not bear an instrumental link to the agent’s ends, then it is no reason at all. The agent sees this, and so issues her denial. This is the explanation that Joyce needs. But a second analysis is just as open. The second analysis is that the agent has a reason after all, though she is absolutely unable to see it. Her denial is based on beliefs she may be justified in holding, but the denial is mistaken nonetheless. The starting points of some agents may be so off-base as to make any flawless reasoning that proceeds from them bound to land in error. If categorical reasons exist, then they are precisely the sorts of considerations to which some agents will remain blind. Since such reasons need not bear any instrumental relation to an agent’s existing motivations, they will, for a certain class of agents, be relevantly unreachable. Agents in such cases will fail to recognize their reasons. Further, nothing these agents can do, when reasoning efficiently from their existing commitments, will take them to the categorical reasons that apply to them.
Error Theory and Normative Ethics
This is a statement of the implications of categorical reasons. It isn’t an argument for their existence. But the argument from action explanation is not an independent argument against categorical reasons. Rather, it operates from an assumption that defenders of categorical reasons should reject. The assumption is that a reason must find a hospitable place within an agent’s motivational set; there must be a link, constructible from one’s commitments, to any genuine reason. But this is precisely what must be argued for.14 Opposing categorical reasons by assuming that reasons must be capable of explaining motivation and action is to assume what must be proved. There is one way that I can think of to defend this basic assumption, though it is not one that Joyce avails himself of, and so he may repudiate it. The path is to explicate the notion of a reason by reference to that of rationality. The relevant argument would take this form, or something like it: 1. Necessarily, something is a reason R for agent S only if a relevantly informed and wholly rational S would be moved by awareness of R. 2. Necessarily, a relevantly informed and wholly rational S would be moved by awareness of R only if R furthered one of S’s ends. 3. Therefore, necessarily, something is a reason R for agent S only if R furthers one of S’s ends.
Given an understanding of categorical reasons as obtaining independently of personal ends, the present conclusion entails that categorical reasons do not exist. For purposes of argument we can accept the instrumental conception of rationality embodied in premise (2). But if we grant that conception, and the premise, then premise (1) is simply question-begging against proponents of categorical reasons. They will deny that perfect instrumental rationality reveals all of one’s reasons. As with logic, so too with reasons: junk in, junk out. Those whose initial commitments are seriously normatively askew cannot be expected to discover the reasons there are for them to refrain from torturing children, to keep one’s word, and to offer aid to the vulnerable. Their ignorance on this score does not immunize them from the reasons that truly apply to them. These last thoughts, admittedly, do not amount to an argument for categorical reasons. They simply report the commitments of those who endorse them. At this point, I just want to point out the seeming coherence and plausibility of these commitments.15 This appearance might be mistaken. But we’ve yet to see a successful argument that shows it so.
Conclusion If a moral error theory is to be vindicated, it must identify the conceptual foundations of morality, and reveal them to be far shakier than we thought. I have tried to show that the best such effort—that of Richard
118 Russ Shafer-Landau
Joyce—is not good enough. It is first of all unclear that categorical reasons are a conceptually nonnegotiable element of morality. Further, even if we are committed in this way, Joyce’s arguments against such reasons are inadequate. Success on either front would be sufficient to undermine the best extant error-theoretic diagnosis of morality. This doesn’t ensure that everything is above-board in our moral practices. We may or may not have an adequate vindication of morality in hand. But it does buy us some breathing room. We don’t—at least as yet—have sufficient reason to regard ourselves as having fallen into error in our basic moral assumptions. If that is so, then the perennial aspirations associated with normative ethical theorizing—that of lighting on the correct conditions of moral assessment—may yet be fulfilled. While ground-clearing of the sort I have undertaken here is hardly the most exciting part of the normative ethical story, I hope at least to have shown that the story is one that might be told, and told well and correctly, perhaps by others within these pages.16
Notes 1. I do my best in Moral Realism: A Defence (Oxford University Press, 2003). 2. Richard Joyce, The Myth of Morality (Cambridge University Press, 2001), ch. 1. Subsequent page references to Joyce are all to this book. 3. An error-theoretic diagnosis is appropriate, says Joyce, for cases in which ‘‘a discourse typically is used in an assertoric manner, but those assertions by and large fail to state truths.’’ (9) Contemporary expressivist works, such as Simon Blackburn’s Ruling Passions (Oxford University Press, 1998), Mark Timmons’ Morality without Foundations (Oxford University Press, 1999) and Allan Gibbard’s Thinking How to Live (Harvard University Press, 2003), seek to retain an assertoric picture of moral talk, and combine it with a minimalism that allows for all the moral truths we want. The aim of these works is therefore to vindicate morality, rather than to debunk it. 4. Ditto for a certain kind of relativism. Perhaps a philosopher such as Gilbert Harman revealed a conceptual misunderstanding when claiming (in ‘‘Moral Relativism Defended,’’ Philosophical Review 84 (1975)) that it made no sense to say of Hitler that he oughtn’t to have ordered the extermination of the Jews. The many critical howls that greeted his claim are evidence that (A) was regarded by his audience as a conceptual truth. 5. I have altered the formulation of Joyce’s argument in a way that I think makes it more forceful. The argument as it appears in his books is as follows: 1. If x morally ought to F, then x ought to F regardless of whether he cares to F, regardless of whether Fing satisfies any of his desires or furthers his interests. 2. If x morally ought to F, then x has a reason for Fing. 3. Therefore, if x morally ought to F, then x has a reason for Fing regardless of whether Fing serves his desires or furthers his interests. 4. But there is no sense to be made of such reasons. 5. Therefore, x is never under a moral obligation. (42).
Error Theory and Normative Ethics
6. 7. 8.
9. 10. 11.
The problem with this formulation is that the consequent in (3) is not, in fact, identical or equivalent to the conjunction of the consequents in (1) and (2). Joyce needs a categorical reason to appear in the consequent of (3); all that follows from (1) and (2), however, is an unqualified reason, conjoined with a categorically applicable ought. Thanks to Tim Hansel for pointing this out. ‘‘Morality as a System of Hypothetical Imperatives,’’ Philosophical Review 81 (1972). Kavka, ‘‘A Reconciliation Project,’’ In D. Copp and D. Zimmerman, Morality, Truth and Reason (Rowman and Allanheld, 1984). Brink, Moral Realism and the Foundations of Ethics (Cambridge University Press, 1989) ch. 3; ‘‘A Puzzle about the Rational Authority of Morality,’’ Philosophical Perspectives 6 (1992). Railton, ‘‘Moral Realism,’’ Philosophical Review 95 (1986); ‘‘Some Questions about the Justification of Morality,’’ Philosophical Perspectives 6 (1992). Copp, Morality, Normativity and Society (Oxford University Press, 1995). To be sure, the central chapters of Joyce’s book are given over to many arguments on the topic, but most are efforts to find holes in arguments for thinking that there are categorical reasons, as offered especially by Michael Smith, in his The Moral Problem (Blackwell, 1994). Since I don’t find those particular arguments very convincing, I will leave them aside. Joyce’s arguments for practical instrumentalism are designed to preempt the need to consider all extant arguments in this area. That’s because if practical instrumentalism is true, then either (i) categorically applicable moral duties would not entail good reasons (thus violating (B)), or (ii) our moral duties would entail good reasons, but only because our moral duties were invariably aligned with our contingent ends. In that case, moral duties would not be categorically applicable (thus violating (A)). Either alternative entails an error theory, according to Joyce. There is a striking similarity between this argument and one offered by Korsgaard, ‘‘Skepticism about Practical Reason,’’ Journal of Philosophy 83 (1986), p. 11. Korsgaard, too, rejects the existence of categorical reasons as construed here. For her, as with all Kantians, reasons must bear a link with our basic commitments. Her disagreement with Joyce (and with Williams, who also argues for an action-explanation requirement on reasons (‘‘Internal and External Reasons,’’ in Moral Luck (Cambridge University Press, 1985), p. 104) is her thought that there are certain ends that all rational agents must have, and that sound reasoning on their basis will generate a set of reasons that applies to all rational agents. Joyce and Williams, by contrast, anticipate a relativism about reasons, a relativism that has as its starting point the assumption that agents can, without rational failure, differ in all of their most basic commitments. Though he might be wrong: it might be rational to remain indifferent to an acknowledged reason, provided one justifiedly believes that the reason is heavily outweighed by competing reasons. There are, to be sure, other arguments in the literature for this thesis, which sometimes goes under the heading of reasons internalism. I survey and criticize what I regard as the most powerful such arguments in Moral Realism: A Defence (Oxford University Press, 2003), ch. 7.
120 Russ Shafer-Landau 15. I provide arguments on behalf of categorical moral reasons in Moral Realism, ch. 8, and in ‘‘A Defense of Categorical Reasons,’’ in progress. 16. I would like to thank audiences at Oxford University, Virginia Tech, Calvin College, Macalester College, and the University of Wisconsin-Milwaukee for many helpful questions and suggestions. Thanks also to my colleagues Dan Hausman, Carolina Sartorio, Juan Comesan˜a, Brie Gertler and Larry Shapiro for an enjoyable session of constructive criticism. Finally, I’d like to thank Brad Majors for the very penetrating comments he provided on an earlier draft.